Effects of inlet boundary conditions, on the computed flow in the Turbine-99 draft tube, using OpenFOAM and CFX

The flow in the Turbine-99 Kaplan draft tube was thoroughly investigated at three workshops (1999, 2001, 2005), which aimed at determining the state of the art of draft tube simulations. The flow is challenging due to the different flow phenomena appearing simultaneously such as unsteadiness, separation, swirl, turbulence, and a strong adverse pressure gradient. The geometry and the experimentally determined inlet boundary conditions were provided to the Turbine-99 workshop participants. At the final workshop, angular resolved inlet velocity boundary conditions were provided. The rotating non-axi-symmetry of the inlet flow due to the runner blades was thus included. The effect of the rotating angular resolution was however not fully investigated at that workshop. The first purpose of this work is to further investigate this effect. Several different inlet boundary conditions are applied – the angular resolved experimental data distributed at the Turbine-99 workshop, the angular resolved results of a runner simulation with interpolated values using different resolution in the tangential and radial directions, and an axi-symmetric variant of the same numerical data. The second purpose of this work is to compare the results from the OpenFOAM and CFX CFD codes, using as similar settings as possible. The present results suggest that the experimental angular inlet boundary conditions proposed to the workshop are not adequate to simulate accurately the flow in the T-99 draft tube. The reason for this is that the experimental phase-averaged data has some important differences compared to the previously measured time-averaged data. Using the interpolated data from the runner simulation as inlet boundary condition however gives good results as long as the resolution of that data is sufficient. It is shown that the difference between the results using the angular-resolved and the corresponding symmetric inlet data is very small, suggesting that the importance of the angular resolution is small. The results from OpenFOAM and CFX are very similar as long as the inlet data resolution is fine enough. CFX seems to be more sensitive to that resolution.